In this paper, we propose a locomotion planning framework for a humanoid robot with stable whole-body collision avoidance motion, which enables the robot to traverse an unknown narrow space on the spot based on environmental measurements. The key idea of the proposed method is to reduce a large computational cost for the whole-body locomotion planning by utilizing global footstep planning results and its centroidal trajectory as a guide. In the global footstep planning phase, we modify the bounding box of the robot approximating the centroidal sway amplitude of the candidate footsteps. This enables the planner to obtain appropriate footsteps and transition time for next whole-body motion planning. Then, we execute sequential whole-body motion planning by prioritized inverse kinematics considering collision avoidance and maintaining its ZMP trajectory, which enables the robot to plan stable motion for each step in 223[Formula: see text]ms at worst. We evaluated the proposed framework by a humanoid robot HRP-5P in the dynamic simulation and the real world. The major contribution of our paper is solving the problem of increasing computational cost for whole-body motion planning and enabling a humanoid robot to execute adaptive on-site locomotion planning in an unknown narrow space.

中文翻译：

---

脚步和质心轨迹引导的具有稳定全身防撞运动的仿人机器人现场运动规划

在本文中，我们提出了一种人形机器人的运动规划框架，该框架具有稳定的全身防撞运动，使机器人能够根据环境测量在现场穿越未知的狭窄空间。该方法的关键思想是通过利用全局足迹规划结果及其质心轨迹作为指导来减少全身运动规划的大量计算成本。在全局脚步规划阶段，我们修改机器人的边界框以近似候选脚步的质心摆动幅度。这使规划者能够为下一次全身运动规划获得适当的脚步和过渡时间。然后，我们通过优先考虑避免碰撞并保持其 ZMP 轨迹的优先逆运动学执行顺序全身运动规划，这使机器人能够在最坏的情况下在 223 [公式：见文本] ms 内规划每个步骤的稳定运动。我们在动态模拟和现实世界中评估了仿人机器人 HRP-5P 提出的框架。我们论文的主要贡献是解决了增加全身运动规划计算成本的问题，并使仿人机器人能够在未知的狭窄空间中执行自适应现场运动规划。